TAG Weekly

4 Sep 2008


See also: IRC log


Stuart, Raman, Ht, Jonathan_Rees, Ashok_Malhotra, DanC.a, TimBL, Dave_Orchard
Noah, Norm


Convene, review agenda/minutes

<skw> Scribe: DanC

PROPOSED: to accept minutes 28 Aug http://www.w3.org/2001/tag/2008/08/28-minutes


thanks to the scribe, Noah Mendelsohn

close action-168

<trackbot> ACTION-168 Circulate draft of our next Quarterly Summary closed

RESOLUTION: to meet again next week 11 Sep; regrets Noah, Norm; ashok to scribe

News/New Items

SKW: how about we do this less frequently... once a month?

AM: as long as we can bring up things that come up

SKW: yes, I solicit input to each week in email

Issue tagSoupIntegration-54 (ISSUE-54)


<trackbot> ACTION-145 -- Tim Berners-Lee to add public prose around his slides at the AC meeting to make the case for extensiblity and flexible XML -- due 2008-06-26 -- OPEN

<trackbot> http://www.w3.org/2001/tag/group/track/actions/145

TimBL: yes, I did some writing and thanks for feedback so far.

[some discussion of interaction between Tim's writing and the TPAC (near NCE) agenda, TAG ftf (near MCI) agenda]

<Zakim> ht, you wanted to query whether TimBL plans to publish before or after TAG f2f

<timbl_tag> Concern oabout spec toglodites

<timbl_tag> (sp?)

<Zakim> raman, you wanted to say that tpac day spent entirely on html and xml would be a good idea as proposed originally by Stuart

<Ashok> Raman, why don't you recommend that

<ht> DC: We are not fixing the holes and moving forward -- IE6 is still very much with us

<ht> HST: I thought Tim's focus was on individual page authors fixing their pages, which I really liked, likewise the bit about giving them good guidance on incremental improvements

<timbl_tag> Exaclty. In HTML we have not been fxing hles and moving forward.

<timbl_tag> But I think in other area we do make teh HTTP a system whcih has got so cruddy and no more , just a 10x blowup from HTTOP 1.0, and then no real changes for 10 years, and a huge amount built i on top

[discussion of consortium priorities and TPAC agenda]

<timbl_tag> I note that we HAVE a fork

<timbl_tag> it is a question as to wheth er we can bring it back to one.

[discussion of TPAC concludes, with emphasis on tagSoup/XML/HTML]


<steve> Heard you were interested in Tech Plenary topics, which is great

<steve> If someone from the TAG would submit their ideas to member-techplenary@w3.org , that would great

<steve> Page with place holder ideas: http://www.w3.org/2008/10/TPAC/TPDay-Agenda.html

<steve> .. also please suggest who would lead each discussion and who might participate

<steve> The program committee would, I'm sure, would appreciate your input. **

<steve> bye

<raman> need to run

[discussion of tagSoup agenda for KC takes the form of something of a preview of the KC discussion...]

Issue abbreviatedURI-56 (ISSUE-56)

SKW: I've seen some email... is it coming to anything?

AM: yes, I think there's a commet the TAG should make

JAR: I'm not sure how deep this XML Schema aspect is.

<timbl_tag> http://www.w3.org/2001/tag/group/track/issues/56

JAR: I think the curie spec should be more clear on the XML types of CURIE and SafeCURIE

TimBL: I have a long-standing action re IRI/URI --

JAR: I'm not concerned by IRI/URI stuff, though there are some editorial things that might be misleading

<Zakim> ht, you wanted to talk about decl for CURIE

HT: w.r.t. both typing and URI/IRI... did my summary message of a few days ago agree with your thinking?

<ht> My message is http://lists.w3.org/Archives/Member/tag/2008Sep/0008.html

JAR: yes

DanC: re whether the XML Schema data type issue is substantive, is it visible from a test case?

JAR: I'm not sure... the "result must be an IRI" constraint can't be checked on the parts independently
... saying "the value space is IRI" is different from xsd;anyURI, whose value space is string

<ht> TimBL, but the lexical space is xxx:yyy, not IRI at all

TimBL: it's coherent for the CURI spec to say the value space is IRI. [?]

<ht> I recommended, in the end "In other words, CURIEs are an abbreviation for strings which are

<ht> _intended_ to represent IRIs, but checking that _intent_ is not part

<ht> of conformance.

<ht> "

JAR: perhaps this will work as a test case... a checker checks datatype well-fomedness...
... if string S passes that check, am I guaranteed that the expansion is an IRI?

TimBL: no

<Zakim> ht, you wanted to say my recommendation implies that it is _not_ an error

JAR: but the CURIE draft says it is, since they say the value space is IRI

TimBL: perhaps I've misunderstood...

<jar> qname is pairs... curie is a single string

HT: mapping from lexical to value is: concat(lookup(prefix), otherpart)
... CURIE mapping, that is...

<noah> FWIW, if I were defining the schema datatype value space for CURIE, it would be the tuple {prefix, suffix, purportedURIstring}, where prefix is the characters before the ":", suffix is the characters after the colon, and purportedURIstring is a string that SHOULD (not must) be a URI

HT: I think the anyURI design is the way to go here: there are no constraints on the value space beyond that it's a string, but the intent is that it's an IRI... and if you try to use it as an IRI and/or URI, problems might happen, but that's outside the scope of the CURIE spec

<noah> Reason for my suggestion: you can imagine software wanting to know what the prefix is, what the suffix is, and of course what the (purported) URI is.

AM: indeed, in practice it's not cost-effective to validate the structure of a URI

<noah> That said, I don't see anything deeply broken in making it like URI, and having only he purportedURIstring as the value.

<DanC_> [jjc's code does it, but indeed, in my experience, his is the only one. and I agree it's not cost-effective to check.]

<ht> jjc's code for which tool does the check?

<DanC_> the RDF validator

<noah> Does it check scheme-specific constraints?

<DanC_> no

<DanC_> even stuff like foo##bar ... in my experience, it works fine to find that error late or not at all

<ht> So, now I realise I'm confused as to who jjc is. . .

<ht> DanC, yes

<ht> Four, I think--Ashok agrees too

<jar> timbl, henry and jar agree: no IRI or URI validation as part of CURIE validation

<jar> ... I think that's what we said ...

HT, JAR, TBL agree that IRI well-formedness should not be a constraint in the CURIE spec

<DanC_> (jjc is Jeremy Carroll as in http://www.w3.org/RDF/Validator/documentation )

SKW: so... how about suggested text?

<ht> Noah has the ball for that, right?

<DanC_> I'd be happy for ht to send his 3-bullet thingy as a trial baloon

<ht> I think we can ask Noah to take my message as a starting point

<DanC_> or you could send it personally, ht

<noah> Yes, I have an action. I'm more or less following the last few minutes. As long as the minutes from this telcon are unambiguous as to what the TAG wants, I can craft a draft response.


<trackbot> ACTION-170 -- Noah Mendelsohn to coordinate response to CURIE last call (with help from Ashok) -- due 2008-09-04 -- OPEN

<trackbot> http://www.w3.org/2001/tag/group/track/actions/170

trackbot, continue action-170

<noah> I think I personally am OK with the direction I believe is being signaled.

<noah> I'll go in and move the date -- I don't think we ever explicitly agreed a date. Note I have sent regrets for 11 Sept, so will not be on the telcon to discuss.

SKW: use TPAC for ftf discussion of CURIEs with XHTML 2 WG?
... I don't hear ftf contact is critical

Responding to WS-*inquiry

DanC: a characature of WS-Transfer is "HTTP over SOAP over HTTP"; I'm trying to learn about contexts where that doesn't seem so silly

<jar> well, we have tcp over tcp, right?

<DanC_> (we do?)

<timbl_tag> We have TCP over SSH over TCP

<timbl_tag> Because the SSH layer provides security '

<noah> Um, well I personally (not with IBM hat on) am a WS-Transfer skeptic, but one context it doesn't look quite so silly is, say, HTTP over SOAP over JMS.

DaveO: if you're a shop that does everything in SOAP, then this is what GET looks like.

<timbl_tag> So the problem is you can't mix soup and soap, because soap is clean and soup is messy.

DanC: this "you do everything using SOAP" view seems different from the "use URIs" view, aka Web Architecture... hmm...

DaveO: indeed, it makes sense to say these are parallel architectures. I've explored bridges and such, to little effect.
... and yes, there is competition at some level... resources in one architecture aren't readily available to the other and such

<timbl_tag> +1 DO

<timbl_tag> "You are locking people into using SOAP and making a whole slew of resources unavailable on the web"

ACTION-171 continues

Issue XMLVersioning-41 (ISSUE-41)

SKW: we're up against time... OK to carry to next week?

DaveO: yes.

<timbl_tag> tx stuart

<jar> bye

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.128 (CVS log)
$Date: 2008/09/04 21:07:51 $